Problem Note 16848: Incorrect output statistics with DIST=MULT and full-rank CLASS coding
Statistics requested in the OUTPUT statement (PREDICTED=, LOWER=,
UPPER=, XBETA=, STDXBETA=) or by the OBSTATS option in the MODEL
statement are incorrect if:
1) the DIST=MULTINOMIAL option is specified in the MODEL statement, and
2) full-rank coding of CLASS variables is specified in the CLASS
statement (PARAM= option). Note that all codings other than GLM are
full-rank.
For models not using the REPEATED statement, you can obtain correct
values by fitting the model in PROC LOGISTIC. This is the default model
fit by PROC LOGISTIC when the response has multiple levels. For an
example, see the Examples section of the LOGISTIC documentation.
Operating System and Release Information
| SAS System | SAS/STAT | Microsoft Windows XP Professional | 9 TS M0 | 9.2 TS1M0 |
| Microsoft Windows NT Workstation | 9 TS M0 | |
| Microsoft® Windows® for 64-Bit Itanium-based Systems | 9 TS M0 | 9.2 TS1M0 |
| Microsoft Windows Server 2003 Standard Edition | 9 TS M0 | 9.2 TS1M0 |
| Microsoft Windows Server 2003 Datacenter Edition | 9 TS M0 | 9.2 TS1M0 |
| Microsoft Windows Server 2003 Enterprise Edition | 9 TS M0 | 9.2 TS1M0 |
| Microsoft Windows 2000 Professional | 9 TS M0 | 9.2 TS1M0 |
| Microsoft Windows 2000 Server | 9 TS M0 | 9.2 TS1M0 |
| Microsoft Windows 2000 Datacenter Server | 9 TS M0 | 9.2 TS1M0 |
| Solaris | 9 TS M0 | 9.2 TS1M0 |
| Microsoft Windows 2000 Advanced Server | 9 TS M0 | 9.2 TS1M0 |
| 64-bit Enabled Solaris | 9 TS M0 | 9.2 TS1M0 |
| 64-bit Enabled AIX | 9 TS M0 | 9.2 TS1M0 |
| OpenVMS Alpha | 9 TS M0 | 9.2 TS1M0 |
| 64-bit Enabled HP-UX | 9 TS M0 | 9.2 TS1M0 |
| HP-UX IPF | 9 TS M0 | 9.2 TS1M0 |
| HP-UX | 9 TS M0 | 9.2 TS1M0 |
| Linux on Itanium | 9.1 TS1M3 | |
| z/OS | 9 TS M0 | 9.2 TS1M0 |
| Linux | 9 TS M0 | 9.2 TS1M0 |
| Tru64 UNIX | 9 TS M0 | 9.2 TS1M0 |
| AIX | 9 TS M0 | 9.2 TS1M0 |
*
For software releases that are not yet generally available, the Fixed
Release is the software release in which the problem is planned to be
fixed.
| Type: | Problem Note |
| Priority: | alert |
| Topic: | Analytics ==> Longitudinal Analysis
Analytics ==> Categorical Data Analysis
SAS Reference ==> Procedures ==> GENMOD
Analytics ==> Analysis of Variance
Analytics ==> Regression
Analytics ==> Bayesian Analysis
|
| Date Modified: | 2006-01-31 09:03:24 |
| Date Created: | 2006-01-12 14:30:47 |
